All issues

The article considers minimization of the expectation of convex function. Problems of this type often arise in machine learning and a variety of other applications. In practice, stochastic gradient descent (SGD) and similar procedures are usually used to solve such problems. We propose to use the ellipsoid method with mini-batching, which converges linearly and can be more efficient than SGD for a class of problems. This is verified by our experiments, which are publicly available. The algorithm does not require neither smoothness nor strong convexity of the objective to achieve linear convergence. Thus, its complexity does not depend on the conditional number of the problem. We prove that the method arrives at an approximate solution with given probability when using mini-batches of size proportional to the desired accuracy to the power −2. This enables efficient parallel execution of the algorithm, whereas possibilities for batch parallelization of SGD are rather limited. Despite fast convergence, ellipsoid method can result in a greater total number of calls to oracle than SGD, which works decently with small batches. Complexity is quadratic in dimension of the problem, hence the method is suitable for relatively small dimensionalities.

Finding the global minimum of a nonconvex function is one of the key and most difficult problems of the modern optimization. In this paper we consider special classes of nonconvex problems which have a clear and distinct global minimum.

In the first part of the paper we consider two classes of «good» nonconvex functions, which can be bounded below and above by a parabolic function. This class of problems has not been widely studied in the literature, although it is rather interesting from an applied point of view. Moreover, for such problems first-order and higher-order methods may be completely ineffective in finding a global minimum. This is due to the fact that the function may oscillate heavily or may be very noisy. Therefore, our new methods use only zero-order information and are based on grid search. The size and fineness of this grid, and hence the guarantee of convergence speed and oracle complexity, depend on the «goodness» of the problem. In particular, we show that if the function is bounded by fairly close parabolic functions, then the complexity is independent of the dimension of the problem. We show that our new methods converge with a linear convergence rate $\log(1/\varepsilon)$ to a global minimum on the cube.

In the second part of the paper, we consider the nonconvex optimization problem from a different angle. We assume that the target minimizing function is the sum of the convex quadratic problem and a nonconvex «noise» function proportional to the distance to the global solution. Considering functions with such noise assumptions for zero-order methods is new in the literature. For such a problem, we use the classical gradient-free approach with gradient approximation through finite differences. We show how the convergence analysis for our problems can be reduced to the standard analysis for convex optimization problems. In particular, we achieve a linear convergence rate for such problems as well.

Experimental results confirm the efficiency and practical applicability of all the obtained methods.

In various applications, i. e., astronomical imaging, electron microscopy, and tomography, images are often damaged by Poisson noise. At the same time, the thermal motion leads to Gaussian noise. Therefore, in such applications, the image is usually corrupted by mixed Poisson – Gaussian noise.

In this paper, we propose a novel method for recovering images corrupted by mixed Poisson – Gaussian noise. In the proposed method, we develop a total variation-based model connected with the nonconvex function and the total generalized variation regularization, which overcomes the staircase artifacts and maintains neat edges.

Numerically, we employ the primal-dual method combined with the classical iteratively reweighted $l_1$ algorithm to solve our minimization problem. Experimental results are provided to demonstrate the superiority of our proposed model and algorithm for mixed Poisson – Gaussian removal to state-of-the-art numerical methods.

An important role in the construction of mathematical models of dynamic systems is played by inverse problems, which in particular include the problem of parametric identification. Unlike classical models that operate with point values, interval models give upper and lower boundaries on the quantities under study. The paper considers an interpolation approach to solving interval problems of parametric identification of dynamic systems for the case when experimental data are represented by external interval estimates. The purpose of the proposed approach is to find such an interval estimate of the model parameters, in which the external interval estimate of the solution of the direct modeling problem would contain experimental data or minimize the deviation from them. The approach is based on the adaptive interpolation algorithm for modeling dynamic systems with interval uncertainties, which makes it possible to explicitly obtain the dependence of phase variables on system parameters. The task of minimizing the distance between the experimental data and the model solution in the space of interval boundaries of the model parameters is formulated. An expression for the gradient of the objectivet function is obtained. On a representative set of tasks, the effectiveness of the proposed approach is demonstrated.

Traditional methods for solving the traveling salesman problem are not effective for high-dimensional problems due to their high computational complexity. One of the most effective ways to solve this problem is the decomposition approach, which includes three main stages: clustering vertices, solving subproblems within each cluster and then merging the obtained solutions into a final solution. This article focuses on the third stage — merging cycles of solving subproblems — since this stage is not always given sufficient attention, which leads to less accurate final solutions of the problem. The paper proposes a new modified Sigal algorithm for merging cycles. To evaluate its effectiveness, it is compared with two algorithms for merging cycles — the method of connecting midpoints of edges and an algorithm based on closeness of cluster centroids. The dependence of quality of solving subproblems on algorithms used for merging cycles is investigated. Sigal’s modified algorithm performs pairwise clustering and minimizes total distance. The centroid method focuses on connecting clusters based on closeness of centroids, and an algorithm using mid-points estimates the distance between mid-points of edges. Two types of clustering — k-means and affinity propagation — were also considered. Numerical experiments were performed using the TSPLIB dataset with different numbers of cities and topologies to test effectiveness of proposed algorithm. The study analyzes errors caused by the order in which clusters were merged, the quality of solving subtasks and number of clusters. Experiments show that the modified Sigal algorithm has the smallest median final distance and the most stable results compared to other methods. Results indicate that the quality of the final solution obtained using the modified Sigal algorithm is more stable depending on the sequence of merging clusters. Improving the quality of solving subproblems usually results in linear improvement of the final solution, but the pooling algorithm rarely affects the degree of this improvement.

The paper proposes and investigates a mathematical model of a distributed computing system of parallel interacting processes competing for the use of a limited number of copies of a structured software resource. In cases of unlimited and limited parallelism by the number of processors of a multiprocessor system, the problems of determining operational and exact values of the execution time of heterogeneous and identically distributed competing processes in a synchronous mode are solved, which ensures a linear order of execution of blocks of a structured software resource within each of the processes without delays. The obtained results can be used in a comparative analysis of mathematical relationships for calculating the implementation time of a set of parallel distributed interacting competing processes, a mathematical study of the efficiency and optimality of the organization of distributed computing, solving problems of constructing an optimal layout of blocks of an identically distributed system, finding the optimal number of processors that provide the directive execution time of given volumes of computations. The proposed models and methods open up new prospects for solving problems of optimal distribution of limited computing resources, synchronization of a set of interacting competing processes, minimization of system costs when executing parallel distributed processes.

Use of algorithms based on neural networks can be inefficient for small amounts of experimental data. Authors consider a solution of this problem in the context of modelling of properties of ceramic composite materials reinforced with carbon nanotubes using perceptron complex. This approach allowed us to obtain a mathematical description of the object of study with a minimal amount of input data (the amount of necessary experimental samples decreased 2–3.3 times). Authors considered different versions of perceptron complex structures. They found that the most appropriate structure has perceptron complex with breakthrough of two input variables. The relative error was only 6%. The selected perceptron complex was shown to be effective for predicting the properties of ceramic composites. The relative errors for output components were 0.3%, 4.2%, 0.4%, 2.9%, and 11.8%.

While modeling seismic wave propagation, it is important to take into account nontrivial topography, as this topography causes multiple complex phenomena, such as diffraction at rough surfaces, complex propagation of Rayleigh waves, and side effects caused by wave interference. The primary goal of this research is to construct a method that implements the free surface on topography, utilizing an overset curved grid for characterization, while keeping the main grid structured rectangular. For a combination of the regular and curve-linear grid, the workability of the grid characteristics method using overset grids (also known as the Chimera grid approach) is analyzed. One of the benefits of this approach is computational complexity reduction, caused by the fact that simulation in a regular, homogeneous physical area using a sparse regular rectangle grid is simpler. The simplification of the mesh building mechanism (one grid is regular, and the other can be automatically built using surface data) is a side effect. Despite its simplicity, the method we propose allows us to increase the digitalization of fractured regions and minimize the Courant number. This paper contains various comparisons of modeling results produced by the proposed method-based solver, and results produced by the well-known solver specfem2d, as well as previous modeling results for the same problems. The drawback of the method is that an interpolation error can worsen an overall model accuracy and reduce the computational schema order. Some countermeasures against it are described. For this paper, only two-dimensional models are analyzed. However, the method we propose can be applied to the three-dimensional problems with minimal adaptation required.

The paper is devoted to convex-concave saddle point problems where the objective is a sum of a large number of functions. Such problems attract considerable attention of the mathematical community due to the variety of applications in machine learning, including adversarial learning, adversarial attacks and robust reinforcement learning, to name a few. The individual functions in the sum usually represent losses related to examples from a data set. Additionally, the formulation admits a possibly nonsmooth composite term. Such terms often reflect regularization in machine learning problems. We assume that the dimension of one of the variable groups is relatively small (about a hundred or less), and the other one is large. This case arises, for example, when one considers the dual formulation for a minimization problem with a moderate number of constraints. The proposed approach is based on using Vaidya’s cutting plane method to minimize with respect to the outer block of variables. This optimization algorithm is especially effective when the dimension of the problem is not very large. An inexact oracle for Vaidya’s method is calculated via an approximate solution of the inner maximization problem, which is solved by the accelerated variance reduced algorithm Katyusha. Thus, we leverage the structure of the problem to achieve fast convergence. Separate complexity bounds for gradients of different components with respect to different variables are obtained in the study. The proposed approach is imposing very mild assumptions about the objective. In particular, neither strong convexity nor smoothness is required with respect to the low-dimensional variable group. The number of steps of the proposed algorithm as well as the arithmetic complexity of each step explicitly depend on the dimensionality of the outer variable, hence the assumption that it is relatively small.

The article focuses on the problem of mechanisms’ co-design for robotic systems to perform adaptive physical interaction with an unstructured environment, including physical human robot interaction. The co-design means simultaneous optimization of mechanics and control system, ensuring optimal behavior and performance of the system. Mechanics optimization refers to the search for optimal structure, geometric parameters, mass distribution among the links and their compliance; control refers to the search for motion trajectories for mechanism’s joints. The paper presents a generalized method of structural-parametric synthesis of underactuated mechanisms with closed kinematics for robotic systems for various purposes, e. g., it was previously used for the co-design of fingers’ mechanisms for anthropomorphic gripper and legs’ mechanisms for galloping robots. The method implements the concept of morphological computation of control laws due to the features of mechanical design, minimizing the control effort from the algorithmic component of the control system, which reduces the requirements for the level of technical equipment and reduces energy consumption. In this paper, the proposed method is used to optimize the structure and geometric parameters of the passive mechanism of the back support module of an industrial exosuit. Human movements are diverse and non-deterministic when compared with the movements of autonomous robots, which complicates the design of wearable robotic devices. To reduce injuries, fatigue and increase the productivity of workers, the synthesized industrial exosuit should not only compensate for loads, but also not interfere with the natural human motions. To test the developed exosuit, kinematic datasets from motion capture of an entire human body during industrial operations were used. The proposed method of structural-parametric synthesis was used to improve the ergonomics of a wearable robotic device. Verification of the synthesized mechanism was carried out using simulation: the passive module of the back is attached to two geometric primitives that move the chest and pelvis of the exosuit operator in accordance with the motion capture data. The ergonomics of the back module is quantified by the distance between the joints connecting the upper and bottom parts of the exosuit; minimizing deviation from the average value corresponds to a lesser limitation of the operator’s movement, i. e. greater ergonomics. The article provides a detailed description of the method of structural-parametric synthesis, an example of synthesis of an exosuit module and the results of simulation.